JPS627967B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention provides differential light scattering for the rapid characterization and identification of large organic cells, particularly white blood cells and other types of mammalian cells.
scattering) techniques. For this purpose, various methods and devices as well as improvements thereof have been proposed. Examples of prior art related to this case include the following. US Patent No.
No. 3624835, No. 3770351, No. 3730842, No.
No. 3754830, No. 3928140, and U.S. Patent Application No. 139366 dated May 3, 1971. There has long been a need for a method to rapidly identify and differentiate large organic particles, particularly white blood cells and other types of mammalian cells. A number of physical techniques have been proposed for this purpose, and various instruments and methods have been developed for using these techniques. Such techniques include chemical staining methods, chromatographic analysis methods, and physical methods such as automated microscopic inspection systems. All of these techniques have been severely limited in their capabilities, the density of cells that can be analyzed, and the speed of analysis. More importantly, the accuracy of many of these techniques has been shown to be questionable. The present invention is directed to methods for rapidly identifying, characterizing and differentiating cells, particularly leukocytes and other mammalian cells. The present invention uses an analytical technique known as differential light scattering. This technique has been shown to be capable of rapid and accurate analysis of microparticles. The terms "identify,""characterize," and "distinguish" cells are used to describe the scope of utility of the present invention. “Identify” means what the cell is (e.g., whether it is a polymorphonuclear white blood cell), whereas “characterize” specifies its physical properties such as size, shape, and dielectric structure. and "identify" is used to separate or differentiate between different types of cells (e.g., distinguishing sickle cells from normal red blood cells, distinguishing normal lymphocytes from abnormal lymphocytes containing contaminants, or (e.g., distinguishing cancerous squamous cells from normal cells). Many publications have described various systems that use differential light scattering techniques to analyze mammalian cell lines. Examples of such publications include Journal of Clinical Chemistry
(Journal of Clinical Chemistry), Volume 21, No.9,
1297-1304 (1975) and Acta Cytolo-gica, Vol. 19, No.
4, pp. 374-377 (1975), there is a paper titled ``Classification of Cells by Laser Light Scattering''. These papers describe systems that use monochromatic light beams to illuminate mammalian cells suspended in water. The resulting differential light scattering pattern is extremely complex. For this reason, there is no empirical method for determining which individual bands of the overall differential light scattering pattern to use to separate a sample population with one property from a population with another property. It was used. Clearly, such empirical methods are severely limited in their applicability to the various cell populations they can handle and their ability to produce meaningful results. The present invention provides methods for rapidly characterizing and identifying cells, particularly white blood cells and other types of mammalian cells. In the method of the present invention,
A polarized monochromatic illumination beam with a wavelength approximately equal to the dimensions of the cells of interest in the cell population to be analyzed is used. Individual cells of the population are illuminated by the beam, and the cells of interest generate differential light scattering (DLS) patterns. This pattern exhibits resonant scattering properties with relatively broad maxima and minima, ie, each maxima and minima extend over an angular range typically between 10 and 30 degrees. These patterns are not so simple in form or so complex in detail that little or no useful information can be derived from them. Rather, because they were generated by a resonant scattering system, these patterns possess meaningful properties sufficient to characterize the individually illuminated scatterers, particularly with respect to physical properties such as their size, shape, and dielectric structure. embodiment, thereby making it possible for scatterers to be precisely and unambiguously identified and identified. The practical application of light scattering phenomena to the identification, characterization and discrimination of large organic particles begins by bringing these particles into liquid suspension in a conventional manner. This suspension is aerosolized and in a preferred method droplets are generated. Some of these droplets will contain a single cell. The droplets are examined and droplets containing cells are separated from other droplets without cells, after which the liquid surrounding each cell is allowed to evaporate, resulting in a free flow of air-bearing cells. Each of these separated cells is sequentially irradiated with a polarized monochromatic radiation beam of wavelength approximately equal to the average size of the cells being examined. For example, for mammalian cell suspensions wavelengths in the infrared region are used. The light scattered by each cell is detected at a sufficient number of angular positions around the cell to provide a differential light scattering pattern unique to the physical properties of that cell. These series of DLS patterns are recorded and analyzed and used to identify, characterize and differentiate cells. As an example of such an analysis, the patterns generated by individual cells can be processed to determine some kind of intensity ratio, such as the ratio of the intensity of a first peak to the intensity of a subsequent peak. Then, cells having a similar number of peaks in the differential light scattering pattern are classified in multidimensional analysis based on these ratios, and cells with similar characteristics are grouped. By preparing a large number of such analysis examples of known cell types of various types, an unknown cell can be identified as one of the known cell types. The detector used to measure the differential light scattering pattern does not need to receive a uniform solid angle or be standardized to a uniform dark stream. By using the same system throughout the analysis, these differences no longer affect the results. from large particles according to the teachings presented here.
Although there are many types of equipment configurations that can appropriately measure, record, and analyze DLS patterns, there are several basic elements needed for a practical system. These include means for handling cells and introducing them into the laser beam one at a time. The laser itself is preferably a plane polarized carbon dioxide infrared source operating at about 10.6 μm, although other sources producing suitable wavelengths may be used. The laser should preferably be coplanar with the line of sight (receive line) of the 10 to 50 individual detector elements. Ideally, the number N of detector elements required in the array is determined by the wavelength λ _p in vacuum, the maximum diameter D of the particles present, the refractive index n _p of the medium in which the measurements are made, and the detector elements. Expressed in terms of the angular range θ that can be covered by the array, the simple formula N[2π
It is given by Dn _p /λ _p +4]θ/180°. For mammalian cells irradiated with 10.6 μm infrared light,
This number is approximately between 10 and 50. Although the detector elements arranged are preferably arranged on an arc spanning 100° or more, the detector elements do not lie on an arc, are not at a distance from the scattering particles, and the measurement angle Satisfactory DLS patterns can also be obtained from other detector element configurations, such as those with ranges less than 100 degrees or those where the detectors are not equidistantly spaced. The present invention will be explained in detail below. In order to understand the present invention, it is important to first understand differential light scattering. Although the basic concept of differential light scattering has been explained in many patents and publications, particularly those by the present inventors, it is briefly stated that it is a method of polarized light scattering to illuminate one or more particles. Use a monochromatic light source. Particles scatter radiation in a manner that is unique to their physical properties such as size, shape, and dielectric properties. The pattern of scattered light is sensed by rotating one collimated detector around the scatterer or using an array of multiple fixed detectors. The measured intensity can be recorded as a function of detector angle to plot the differential light scattering pattern. Such differential light scattering patterns may contain a great deal of information about the scatterer, or they may provide little to no information about the scatterer. For example, if the size of the particle is very small relative to the wavelength of the illumination beam, little to no change will be exhibited in the intensity of the scattered radiation as a function of the measurement angle around the particle. On the other hand,
If the particle size is very large relative to the wavelength of the illumination beam, the scattering pattern will exhibit a large number of maxima and minima. Interpreting such patterns in order to derive significant particle characteristics from them is a major challenge. This analysis task can be greatly simplified by adjusting the wavelength of the illuminating monochromatic light source to be approximately equal to the overall size of the irradiated particles. When such approximate equivalence exists, i.e. when the particle is in the resonance region with the illuminating light, the resulting differential light scattering pattern still exhibits maxima and minima, but many features that are directly correlated to the structure of the particle It is not so complicated that the analysis work would be a waste of effort because it is expressed in such detail. The use of resonant scattering techniques is highly significant.
The differential light scattering pattern is a complex function of the size, shape, orientation and structure of the particles and the polarization and wavelength of the incident radiation. The most important scattering parameter is the normalized dimension, ρ=πDn _p /λ _p =ka (1) where a: average particle radius D (=2a): average particle diameter λ _p : vacuum wavelength n of the incident beam _p : refractive index of the medium surrounding the particle. Although variations in the dimension parameter ρ have a large effect on the corresponding differential light scattering pattern, the dimension itself is important for distinguishing abnormal cells from normal cells,
As a parameter for distinguishing one type of red blood cell from another, or one type of pollen from another, it is one of the least important and most ambiguous parameters. In addition, the size distributions are permanently overlapping. As ρ becomes very large, the differential light scattering pattern becomes full of additional peaks of little analytical significance. If a DLS pattern is recorded by an array of N detector elements, the optimal number of these detectors arranged over the entire angular range from 0° to 180 is simply given by : N2ρ+4 (2) In other words, from intensity data recorded at N well-spaced positions over a 180° range, the DLS pattern between these positions can be interpolated with great accuracy. . The angular range of interest is this
If it is less than 180°, then the number of detector elements can be reduced accordingly by the ratio of that angular range to 180°. Incidentally, for a particular measurement using such an array to sense a DLS pattern, the maximum number of detector elements required will depend on the size of the largest grain of interest to be measured. fully specified by Thus, N should always be chosen to be approximately 2ρ _nax +4, where ρ _nax is the expected normalized size of the largest of the grains of interest in the suspension being examined. Average diameter illuminated by infrared wavelength of 10.6μm
For a 60 μm airborne scale cell, the optimal number of detector elements is approximately 2π60/10.6+440 (3). At large scattering angles, internal organization or structure plays an important role in determining changes in scattering properties. In fact, this fact was previously pointed out by the inventor (Applied Optics Vol. 7, pp. 1879-1896, 1968).
Year). This fact is confirmed by the Los Alamos Scientific Laboratory.
This became the basis for extensive differential light scattering research conducted at the Laboratory. It was hoped that by using even larger angle differential light scattering measurements, it would be possible to differentiate between similar types of mammalian cells. These measurements were made using a He--Ne laser operating at a visible wavelength of 632.8 nm. Such wavelengths are extremely small compared to the average size of the scale cells being examined. Although some results have been reported, these measurements using such short wavelengths have reached their ultimate limit and no further progress can be made. The reason for this is obvious. At large scattering angles, the differential light scattering pattern for such large values of ρ depends closely on ρ through the following parameters: ξ = ρ (4) (Here it is the average refractive index of the scattering particles divided by n _p .) Since different types of mammalian cells are expected to have slightly different values ( Some degree of separation based on the ρ measurements is to be expected (although importantly they fall into domains of different dimensions). Although the reported distinction using "cluster algorithms" is fortuitous, it would be almost impossible to distinguish between cells with comparable ρ values. Although he was aware of the importance of using long wavelength radiation, problems associated with infrared radiation prevented him from even attempting it. The present invention provides that the identification and characterization of mammalian cells and other large organic grains by differential light scattering techniques can be accomplished using infrared radiation at these resonant wavelengths of the irradiated grains. This is based on the recognition that the dimensions are practical because they correspond to the wavelength of the irradiating radiation. This may be explained by considering the two remaining dimensional relationships: when the particles are very small compared to the irradiating radiation, and when the particles are very large compared to the irradiating radiation. In the former case, for example using microwave radiation, it can be shown that there is no angle-dependent variation in the scattering intensity. Therefore, intensity measurements only need to be made at one arbitrary angle. The absolute scattered intensity from a single small particle cannot be used to classify or separate particles, even if it can actually be measured. This is because such measurands are ambiguous. ie, a single intensity measurement cannot be associated with a particular class of mammalian cells. A slightly different normalized dimension, ρ, is easily compensated for by a variation in the average relative refractive index, and vice versa. Moreover, slight structural or geometrical differences between different types of cells have no single measured effect on the scattered intensity, and therefore these differences are also undetectable. In the latter case, for example using visible light, when the grain dimensions are very large compared to the wavelength of the illuminating radiation, the differential light scattering pattern will scatter more than is needed to explain the important structural properties of the grain. Excessive details of light. An example of such patterns and the intricate details they exhibit can be found in Applied Optics Vol. 9, (1970).
This is shown in the paper "Scattering by Individual Transparent Spheres" published from page 2522 onwards. The plethora of data for each measurement presented therein comes from a very simple scattering device: a spherical droplet of uniform dielectric structure is exposed to polarized monochromatic radiation at a wavelength of approximately 1/100 of the particle size. irradiated. The present invention is intended for the rapid identification, discrimination, and characterization of cells that are extremely large compared to the wavelengths of visible light, particularly mammalian cells of various types. These cells often contain a membrane of one dielectric tissue surrounding an interior of another dielectric tissue, and these may further contain various grains or anomalies of another dielectric tissue. There is. The shapes and dimensions of these various tissues are often non-circular, and some are platelet-like, for example. If visible light irradiation, i.e. radiation whose wavelength is very small compared to the average diameter of these particles, were used to obtain a differential light scattering pattern for such particles, a very large number of maxima and minima would occur in this pattern. In addition to appearing redundantly, even small changes in some of these tissue properties result in very large changes in the scattering pattern. The foregoing indicates that attempts to use such complex differential light scattering patterns to discriminate between grains with different compositions are near futile. As previously mentioned, the present invention provides a method for rapidly identifying and differentiating large organic cells, particularly mammalian cells. Such cells typically range in diameter from a few micrometers to around several tens of micrometers. These cells include red blood cells, white blood cells and scaly cells. For example 10 such individual cells
When illuminated with a polarized monochromatic light beam of wavelength on the order of μm, broad maxima and minima appear in the differential light scattering pattern produced by this system. This relationship, where the grain size is in the so-called resonance region, is
It produces patterns with properties that are much more easily correlated to structural differences in the scattered object. Moderate changes in the average properties of the grains result in noticeable changes in the pattern over a certain angular range, but for grains of comparable size the overall pattern is very similar in shape. Moreover, such patterns contain sufficient detail to allow accurate characterization and identification of various grains, but do not contain enough detail to obscure or preclude accurate mathematical interpretation of the physical properties of the scattered objects. does not contain a lot of unnecessary information. The recognition that the wavelength of the irradiation beam and the irradiated grain should have approximately equal size makes practical the rapid and unambiguous characterization and identification of mammalian cells and other large organic cells. It also provides a relatively simple yet highly practical method for accomplishing such analysis. The need for this relationship dictates the use of infrared lasers for the analysis of mammalian cell lines. Mammalian cells contain and are usually surrounded by an aqueous fluid. This presents significant difficulties. The absorption number of water is very large in the infrared region. Thus, water, which is normally present within or surrounding cells, plays an important role in resonant differential light scattering measurements from mammalian cells.
When an aqueous suspension of cells is irradiated with infrared radiation, one would expect attenuation of the irradiation beam on the order of 90% or more within a distance as small as the dimensions of the cells. Moreover, such attenuation would adversely affect the radiation scattered by the cells by severely distorting and attenuating the scattered waves emerging from the suspension. Therefore, it is not surprising that the field of differential light scattering and infrared measurements had long ago given up on the possibility of making meaningful DLS measurements from such large organic particles at these wavelengths. do not have. However, the well-established idea that attenuation by water within the water envelope or the air-bearing cells themselves precludes the practical application of infrared radiation appears to be spurious. Recognizing that scattering of radiation by objects in the resonance region is generally not governed by geometric-optic attenuation relationships, expected scattering properties were measured for water-supported particles in the infrared resonance region. These show conclusively that such particles can be identified by their DLS patterns if they are surrounded by a gas such as air that does not strongly absorb infrared radiation. For this reason, it is preferable to accomplish cell scattering measurements in a gaseous atmosphere. Although air is used as the gas in the preferred configuration just described, it should be understood that any of a variety of other gases may be used if desired. Furthermore, analysis can also be accomplished in vacuum if the grains of interest can maintain their texture in a vacuum atmosphere. Also, if an infrared transparent fluid is found, measurements can also be accomplished in such an atmosphere. If only a very thin layer of water coats the cells, measurements can be easily achieved. As previously mentioned, the practical application of resonant scattering phenomena to relatively large organic particles such as mammalian cells or pollen particles requires the use of polarized infrared radiation at wavelengths of about 10 μm. In summary, cells or particles are first aerosolized and then transported through the detector array by a laminar flow of drying gas. A detector array consisting of N(2ρ _nax +4) individual detector elements is:
As the individual particles pass through the collimated laser beam, they experience a DLS pattern with an angular range of approximately 100° generated by the individual particles. The laser beam is preferably coplanar with and perpendicular to the detector array element. Some detectors must be refrigerant cooled and isolated from the dry gas flow by an insulating cylinder containing a window made of an infrared transparent material such as germanium. As the grain passes through the laser beam, it passes through the period D/
A spherical wave pulse of V (where D is the beam diameter and V is the particle/flow velocity) is generated. The individual detector elements convert the pulses they receive into electrical signals, which are then intensified and preferably stored as digital signals in computer memory or on tape. These N2ρ _nax +4 stored signals can be used to reconstruct continuous DLS patterns from grains by interpolation using Tievisieb polynomials or related techniques, or the stored signals can also be used directly. The angular spacing of the individual elements may be equidistant. However, for some types of subsequent analysis, the Tievisieb polynomial T _N-
It would be preferable to space _the detectors according to the N-root position of ₁ (X), where N = number _of detector elements, (θ ₂ +
θ ₁ )/2], θ ₁ and θ ₂ (θ ₁ <θ ₂ ) are
(defines the angular range over which the DLS pattern is recorded). Once the DLS patterns, or the array data points from which such patterns are reconstructed, have been stored, these patterns can be processed by various algorithms so that each grain can be identified or otherwise characterized accordingly. can be analyzed. It has been found that a most practical algorithm is based on a characterization method that first classifies the grains into sets of equal size. The average grain size is between the two angular limits
It can be adequately evaluated by counting the number of DLS peaks or by measuring the angular position of a particular peak in the forward direction. Since such inferred size parameters are not a definitive means for identification or discrimination, their exact values are not important as long as grains of the same effective size are compared with each other. all of the grains in a given set have effectively the same dimensions,
For example, grains can be structurally and physically distinguished from each other by various classification algorithms if they are found to have the same number of peaks in two angular limits. These algorithms are given DLS
DLS by comparing various ratios of patterns
Patterns can be characterized. For example, the ratio of the heights of several DLS peaks in a given angular range to the height of the first peak in that range is a useful set of discrimination parameters, as detailed below. Other ratios include ratios associated with functionally complex displays based on different peak-to-trough values and the different peak heights and valley depths present. All such ratios are a function of the dielectric properties of the scattering grains and can be used to characterize each grain. Once such ratio characterizations have been realized for each of the constant size groups present, they are further processed by a storage catalog of these ratios stored in the computer memory of the system's analytical processor. Parsed. The distribution of such scattering pattern ratios as a function of grain size represents another important means for identifying, characterizing, and discriminating grains. As shown in Figure 1, the suspension of cells to be analyzed was found in the article by WABonner et al. published in Rev.Sci.Instruments, pp. 43, 404 et seq. A cell sorter 2 similar to that described in "Fluorescent Activated Cell Sorting" is fed to a cell sorter 2 similar to that described in "Fluorescent Activated Cell Sorting". The sorter breaks up a liquid suspension of cells into a group of individual droplets, each droplet being of sufficiently small size to not contain more than one cell or particle. As explained in detail later in Figure 2,
These droplets are electrostatically charged, illuminated by a beam of light to determine whether they contain particles, and then sorted by electrostatic deflection to produce a droplet stream containing only cell-bearing droplets. Ru. This droplet stream 4 is fed to a detector system 6. The detector system illuminates each droplet with a polarized monochromatic beam, preferably an infrared laser at a wavelength of 10.6 micrometers. The scattered light sequentially scattered by each cell is measured by a sensor system, preferably a group of sensor arrays incorporated within the detector 6. The output electrical signal from the detector is transmitted into a recording device 8, such as a magnetic tape recorder, and stored therein. The stored data is then analyzed by a computer analyzer. After sufficient particles have been analyzed, a computer compiles the results and provides output in the form of tape, disk, or hard copy, as described below. FIG. 2 illustrates details of the sorter and detector.
As described in the above-mentioned article, a liquid suspension of cells passes through an orifice 22 in a tube 24 to produce a series of droplets 26. Each droplet is charged by an ionization source. Individual droplets are illuminated by beam 28 and the scattering intensity from each droplet is sensed by detector 30, amplified and transmitted to analyzer 32. Analyzer 32 provides an electrical potential to a set of electrostatic deflection plates 34 in a controlled manner. The droplet stream passes between the plates of this electrostatic deflection device. The scattering generated by successively falling droplets and measured by detector 30 is analyzed to determine the presence or absence of cells within the droplet. Depending on the results of the analysis, the analyzer energizes a set of deflection plates to electrostatically deflect cell-free droplets 36 from the flow. The remaining droplet 38, ie the droplet containing the cells or particles, passes to the detector system. The detector system includes a housing or scattering chamber 4
2. Sensor array 44, inlet 46 and outlet 4
Contains 8. A radiation source 50, preferably a laser, generates a monochromatic beam of light, preferably plane polarized infrared light 52. The beam enters housing 42 through a window 56 in an aperture 54 thereof and passes through the housing to a Rayleigh horn or light trap 58. Preferably, the axis through which the beam passes lies in the plane of the detector array 44 and is perpendicular to the droplet flow down axis. However, as will be seen later, this relationship is not an essential requirement for the operation of the present device. The detector array is preferably a liquid nitrogen cooled multi-element mercury-cadmium telluride [HgCd
(Te)]. It is manufactured by HoneywellCorp. Radiation Center. Other suitable detectors are manufactured by Arthur D. Little & Co. and the Santa Barbara Research Center, a division of Hughes Aircraft. These detectors have very high detection capabilities and are ideal for measurements such as those in the present invention. However, having to cool them to low temperatures is inconvenient and undesirable for some applications. Therefore, pyroelectric detectors that can operate at room temperature are also suitable for measurements near 10 μm. The detection power of such a detector is
It is about 100 times lower than that of the Hg-Cd(Te) type, but as radiation sources with almost unlimited power, such as CO ₂ lasers, are becoming available, adequate scattering can be achieved. It is guaranteed that a signal will be obtained. Pyroelectric detectors are significantly cheaper than Hg-Cd(Te) detectors, thereby allowing for significant reductions in manufacturing and operating costs. The Honeywell Radiation Center has published a collection of papers entitled ``Summary of Honeywell Publications on Pyroelectric Detectors and Materials,'' which presents a variety of sensing devices. For example, the following two are noteworthy: Ferro Electrics (Ferro Electrics)
electrics 3 , pp. 281-285 (1972) "Pyroelectric properties of lanthanum-doped ferroelectric PLZT ceramics"; Physica 61 , pp. 589-593 (1972) "Pyroelectric radiation detectors" ``Noise sources in''. Preferably,
The individual detector elements of the array are approximately 2 mm apart from each other, and 10 to 50 individual elements are distributed along an arc spanning approximately 100° from a scattering angle of 30° to a scattering angle of 130°. be done. If the element is a Hg-Cd (Te) type detector, such an array must be cooled with a refrigerant. For this purpose, a liquid nitrogen source (Hg−Cd(Te) of approximately 77
) is supplied to a jacket 62 (FIG. 3) which is incorporated within the housing 42 and surrounding the detector array 44. Preferably, the inner surfaces of the circles circumferentially spacing the arrays have a radius of about 1 cm. If the detectors are cooled with a refrigerant, they must be isolated from the atmospheric atmosphere by interposing a vacuum between the detector and the circulating atmospheric flow. This is most easily accomplished with a concentric inner structure 64, in which case the area between the array 44 and the inner structure 64 is evacuated to a vacuum. The inner structure includes a suitable window made of germanium or any other infrared transparent material. The liquid portion of the droplet surrounding the cells evaporates as the cells pass from the cell classifier into and through the detector. Preferably, the humidity of the transfer air stream is adjusted such that the liquid containing the cells in the droplet evaporates just during the transfer of the cells from the classifier to the detector, so that the free air-bearing cells are irradiated in the detector zone. be done. Thus, the atmosphere flowing with the cells tends to be rather humid. Should this humid atmosphere encounter refrigerant temperatures or be cooled to a significant degree, condensation is likely to occur, and such condensation would significantly adversely affect light scattering measurements. To avoid this problem, it is preferred to minimize the volume of atmosphere flowing into the detector with the cell flow, while simultaneously surrounding this atmosphere and the cell flow with an insulating envelope layer of dry gas.
Also, as mentioned above, the aerosolization process should be sufficiently isolated from the detector so that any adhered water is evaporated without dehydrating the cells. Ideally, the liquid surrounding each cell would be evaporated just before the cell passes through the detector array, and the liquid inside the cell would not evaporate through the cell membrane. The dry air outer layer is a dry air source 6
9 through collar 66 and which introduces dry air as a laminar flow column around the cell flow. This column of dry air isolates the cell flow from the infrared window 56. The flow of cells and the surrounding columnar flow of drying gas are discharged through a discharge opening 48 in the housing and through a conduit 68 at negative pressure into a container (not shown). As clearly shown in FIG. 3, as individual cells pass through the radiation beam 52, they cause the beam to scatter. Some of the scattered light is transmitted to the detector array 4
4 are intercepted by individual detectors at 4. The individual electrical signals produced by each detector in the array are transmitted through cable 88 to a subsequent electronic analysis and recording system. Particularly when considering mammalian cells, if the data contain significant structural and surface differences in addition to dimensional and refractive index changes, the data are amenable to analysis well beyond the current level of analytical techniques. Let us present the task of interpretation. This situation can easily occur when large particles are irradiated with visible light. Most importantly, the information overload of details in such differential light scattering patterns produced using visible light appears to be so fine that the physical data that can be inferred from them is no longer interpretable. It can be done. Extracting only the salient grain characteristics from such a pattern, while possible, is at best a laborious undertaking. However, an argument may be made that differential light scattering patterns could be recorded with much less angular resolution, thereby minimizing the amount of data for the physical parameters involved. However, as revealed by the study of patterns such as those presented in Applied Optics by Blau et al. mentioned above, the envelope of such a differential light scattering pattern is characterized by small changes in the structure of the scattering particles. It also fluctuates significantly. Thus, reducing the angular decomposition does not in itself result in a satisfactory improvement of the parameter interpretation inference problem. Figures 4A and B show a vertically polarized beam of monochromatic radiation with a wavelength of 10.6 μm (with an electric vector in a plane perpendicular to the plane measured by the sensor).
A computer-generated differential light scattering pattern is illustrated for cells with high water content illuminated by . The cells in Figure 4A have a radius of 10 μm and the cells in Figure 4B have a radius of 20 μm. The dimensions of the former will be similar to white blood cells, while the latter will be closer to those of scaly cells. In these examples, 4
Two different compositions were selected, ie, different compositions in which the amount of protein in the particles was slightly different. Since the presence of additional protein increases only the real part of the refractive index of the particles, the four selected examples exhibit negligible changes in the real part of their average refractive index: A: n=1.176+i0. 084 (This is approximately equal to the refractive index of 10.6 μm pure water, i.e., an empty droplet) B: n = 1.20 + i0.084 C: n = 1.25 + i0.084 D: n = 1.30 + i0.084 The curves are in the order indicated by their respective letters. Figures 4C and 4
Figure D is identical to those given in Figures 4A and 4B, except that their patterns are obtained from horizontal polarization of the illuminating infrared radiation (with the electric vector in the plane coinciding with the plane measured by the detector). An example of a pattern corresponding to the same particle is shown below. As this data illustrates, relatively simple structural changes in the scattering particles result in relatively simple changes in the differential light scattering pattern. For small particles, the horizontally polarized light scattering intensity at about 50° versus its intensity at about 40° provides a simple yet elegant means of this postulated cellular physical difference. A similar simple and obvious difference for large cells is shown in Figure 4D. Therefore, the differential light scattering patterns produced by such cells when irradiated using short wavelength visible light are extremely complex and the number of physical parameters involved gives rise to a data overload; When the particles and radiation are in the resonance region, the corresponding patterns obtained at long infrared wavelengths are much easier to understand and produce data that are easier to interpret.
This allows particles to be accurately identified, differentiated and analyzed using relatively simple and readily available computerized equipment. It should be particularly noted that the particles producing the patterns illustrated in Figures 4A-D have a refractive index very close to that of water in the infrared, i.e. have a high water content and are therefore very absorbent in the infrared. be.
Nevertheless, their scattering patterns show distinct differences for relatively small but significant compositional changes. Due to the large imaginary part of the refractive index, such differences are generally not expected. Therefore, until now, those skilled in the art have avoided the use of infrared radiation and have not even considered the intuitively incorrect predictions described above. A preferred analysis system is shown in FIG. As previously explained, as each cell of the cell suspension passes through the cell classifier, each cell produces a differential light scattering pattern that is sensed by detector array 44. In one embodiment of a system such as that shown in FIG. It is recorded by recorders 8 and each recorder provides one channel for each detector in the array. The intensity variations that occur in response to the recorded output are then first analyzed to determine the maximum intensity of each detector for each cell being analyzed. The maximum intensities are combined to create a differential light scattering pattern as shown in FIG. These patterns, or portions thereof, can be analyzed by any of a variety of methods to identify, classify, and characterize the cells that gave rise to them. For example, referring to Figure 4B, particles of similar size but with slightly different refractive indices depending on changes in protein composition show that the magnitude of the second peak relative to the magnitude of the first peak (approximately 25°) can be distinguished from each other based on their The table below summarizes these ratios for the first four peaks.

Table: Based on these different peak ratios, particles of the same size but with slightly different physical properties can be easily distinguished, as this table easily shows. Thus. To distinguish between particles of different sizes and different textures, such as those present in a suspension of mammalian cells, the differential light scattering patterns produced by the various individual particles are first compared to the peaks they exhibit. This sorting operation divides the resulting patterns into sets of scattering patterns originating from particles of approximately the same size. After such rough size classification,
Particles of approximately the same size are compared to each other. As a first identification of these particles, the intensity of the first peak is used as a standard value and the second peak relative to this first peak is used as a standard value.
The ratio of peak intensities is used as a more accurate classification than simply size classification. In fact, as mentioned above, leukocytes contain a number of different cell types, and these various types of cells have a size of approximately 8 μm.
They range in size from lymphocytes at around 18 μm to granulocytes at approximately 18 μm. The dimensions of a significant number of these cells are approximately equal, for example on the order of 12-13 μm. Such particles may have about 4-5 peaks in their light scattering pattern when irradiated with vertically polarized monochromatic infrared light. If the ratio of the first peak to the second peak is used to identify these particles in a simple two-dimensional analysis, a variety of overlapping Gaussian distributions will generally result in overlapping distributions of leukocyte types in this narrowed size range. therefore brought about. To further differentiate these distributions, each successive peak ratio can be used in a multidimensional vector analysis. Although such tasks can be accomplished manually, for example, a typical multidimensional vector space partitioning analysis for classifying samples with similar characteristics uses an appropriately programmed electronic computer. It is more convenient to use standard pattern recognition techniques such as . Such analysis techniques are well known to those skilled in the art and are routinely performed today to classify complex data using multidimensional arrays. When a large number of cells are to be analyzed or when for some reason it is inconvenient to accomplish cell identification and classification manually, an electronic system such as that shown in FIG. 5 may be used. In this system, the output of each detector in detector array 44 is preferably provided to a logarithmic amplifier 72. By converting each detector output signal, which is a linear function of the light intensity generated at each detector, into a logarithmic value, as accomplished by a logarithmic amplifier 72,
The dynamic range of the system is increased considerably without increasing digital data processing requirements. In addition,
For example, comparing ratios requires simply subtracting logarithmic values, which greatly simplifies data processing and comparison. On the other hand, with the rapid emergence of inexpensive digital computers, the use of linear amplifiers to additionally handle more complex computational tasks is equally attractive. The response of these separate logarithmic to linear amplifiers is standardized by first projecting light of uniform intensity onto all of the detectors in the array simultaneously and then adjusting the amplifiers so that all produce the same output. It can be done. Detector standardization is not required if the absolute difference between the gains of individual detectors is measured and stored for subsequent calculation correction. Alternatively, the intensity setting generated by a single particle can be used as a reference setting and thereby standardize or modify all subsequent settings. Because of the excellent linearity of each detector in the infrared-sensitive detector array discussed earlier, the output of the logarithmic amplifier, after standardization adjustment, represents the logarithm of each intensity of radiation impinging on each detector. Represent accurately. These logarithmic amplifiers are manufactured by Analog Devices, Inc. and are manufactured by Device No.
755 can be used. These respective outputs are transmitted to sample and peak detector 74. The peak detector is, for example, Buyul Braun equipment No. 4084.
It can be done. A discriminator 76 is coupled to a logarithmic amplifier that provides the output of the smallest angle detector. As the intensity generated by this low angle detector changes in response to the passage of the particle, this change is recorded by the discriminator. Discriminator 76 is also connected to peak detector 74 and holds them in the clear state until the previous analysis cycle is completed and the next cycle is initiated. This is triggered by the intensity of the output of a logarithmic amplifier connected to the detector at the lowest angle above a predetermined level sufficient to indicate that a particle is passing through the beam of monochromatic radiation. As the particles pass through the beam, the output of each detector changes until it reaches a maximum value, which is stored by the peak detector 74 connected to it. These stored intensities correspond to the intensities of the differential light scattering patterns produced by the particle at various sequential angles of the detector. As the particle passes out of the laser beam, the intensity sensed by the lowest angle detector diminishes and disappears. Discriminator 76 responds to this decreasing intensity and activates control logic 78 via connection line 80. Control logic circuit 78
ultimately operates the analog-to-digital converter 82. The A/D converter 82 has each peak detector 74
are sequentially connected to each other by a multiplexer 84.
Such multiplexers and converters are available, for example, from Buyul Braun's Data Acquisition Unit.
Available as MP8126. As a result of this processing, the logarithmized analog signal stored in each of the peak detectors is converted to a digital representation. The digital representations are conveyed to a storage device 86, preferably formed by emitter-coupled logic elements such as those manufactured by Motorola, where they are sequentially stored. Thus, stored in the memory is a digital representation of the maximum value of the scattered light intensity sensed by each successive detector in array 44. After this operation, control logic 78 signals discriminator 76 to allow it to accept the next new data. In the preferred embodiment, memory 86 is connected to microprocessor 88. The microprocessor examines the data by traversing the digital information stored in memory to determine the number of peaks present, and if the number and spacing of detectors is insufficient to give the desired accuracy, a mathematical Use interpolation. This study aims to determine the number of such peaks,
Provides a digital sequence output representing position and value. More specifically, the microprocessor analyzes the data to produce a first ratio value, e.g., to determine the ratio of the measured second peak intensity to the measured first peak intensity; A digital representation is maintained by a microprocessor. In a similar manner, the microprocessor processes the data stored in the memory to determine the successive peak ratios, thereby first determining the differential light produced by the grain just sensed by the detector array. A digital output is produced indicating the number of peaks in the scattering pattern and then some peak ratio values for this particle, such as the ratio values presented in the table. Microprocessor 88 and control logic circuit 78
Both may use bipolar high speed bit slicing microprocessors manufactured by Motorola or Texas Instruments (eg, Motorola Microprocessor No. MC10800).
The microprocessor is controlled by a programmable read-only memory to accomplish the sequential analysis described above or any other desired analysis. The resulting digital data stream is stored on a disk-type data storage unit 90, displayed on a video terminal 92 or printed on a printer 94.
is output as a hard copy output. Alternatively, it may be stored in a large storage device.
Storage, video terminals and printers can be connected directly to the microprocessor, but preferably further analysis of the data is accomplished by minicomputer 96. The central processing unit first causes the data to be transmitted to disk data storage unit 90. The processing unit then analyzes the stored data, e.g. by a multidimensional vector space partitioning analysis program or other suitable sorting algorithm as previously described, to determine the various types of suspension present in the suspension supplied to the system. A video representation of the cell type is displayed on terminal 92, and this representation is printed by printer 94 in response to user commands. The minicomputer 96 is, for example, a PDP11-20 manufactured by Digital Equipment. Of course, various other computer systems could accomplish this analysis quite satisfactorily. Many conventional systems that use a detector or detector array to measure light scattered by an object along a substantial arc of a circle move the detector or detector array at a constant radius from the object throughout the measuring arc. It emphasizes the importance of maintaining distance. As mentioned above, it is preferred to use a detector array in the method of the invention. This requirement to maintain a constant radial distance imposes significant limitations on the array. Not only must the device be constructed to form an arc of suitable radius and length, but according to conventional teachings, the sensitivity of each element of the detector array must be very uniform. Such limitations significantly increase the cost of the detector array and associated electronic systems required to achieve and maintain such uniformity. The reason for this requirement is that the light intensity decreases inversely as the square of the distance from the scattering object. Thus, if a detector array is used and all of the detectors in the array are not uniformly at the same distance from the object emitting radiation, the light intensities impinging on the elements of the array will not be equal. Furthermore, the surface areas of the elements must be exactly the same so that they receive the same solid angle of radiation. All of this is to achieve uniformity of response of each detector in the array to ensure uniform scattering by the illuminated object. Only when such uniformity is achieved can a light scattering pattern as illustrated in FIG. 4 be achieved. An important aspect of the invention is that such uniformity need not exist in the detector array. In practice, the detector array may consist of a number of linear sections arranged along the interior of the housing. Such a linear section is shaped as shown in FIG. Of course, more or fewer compartments can be used as desired, and they can be configured in a variety of other ways. Each adjacent element or detector of the array at a different radial distance from the scattering object receives scattered light at a different solid angle. Additionally, these detectors do not need to be in the same plane. These differences in the array result in significant distortions of the intensity of light sensed by the detectors that make up the array. This distortion can be thought of as a simple modification of the undistorted scattering pattern. However, such a modification does not lead to incorrect characterization of the analyzed cells. Light scattered by each substantially equivalent cell results in a distorted but substantially equivalent scattering pattern that is provided to a data processing device. Similarly, cells of different characteristics result in correspondingly different scattering patterns that are similarly deformed, which are fed to the data processing device. identification,
For characterization or identification purposes, consistency is established between the array elements and between the responses transmitted to the data processing system from equivalent particles irradiated upon passage through the detector housing, and within the detector housing. It is only necessary to establish a difference between the modified scattering patterns applied to the preprocessor for the different cells being irradiated. Although a detector arrangement composed of various linear detector sections, such as that shown in FIG. cells are irradiated, resulting in substantially identical light scattering patterns being provided to the preprocessor, and substantially different cells resulting in substantially different patterns being provided to the processing system. . Thus, the system is still able to correlate cells that are substantially equivalent and to discriminate between cells that are not equivalent.
For this reason, considerable savings in the cost and simplification of the structure of the detector array are achieved, which is the main objective of the present invention, namely the rapid detection of mammalian cells, pollen and other large particles such as fungal plant spores. This is achieved in achieving unambiguous identification, characterization and identification. It should be noted that the modified DLS patterns measured for particle identification and discrimination can be of many different types and measured in many different ways. Although a detector array is preferred for this purpose, it is not mandatory. Other methods exist for measuring and recording DLS patterns. If the illumination source is coaxial with the particle stream, a single detector can synthesize a continuous array of DLS patterns. Such a configuration is illustrated in FIG. The radiation beam 102 is reflected by a mirror 104 dimensioned to direct the beam along the falling path of the cells 106. As each cell enters the detector housing 108, the cell enters the light receiving field of a single detector 110. During the migration of the cells through the housing, the scattered light from the cells is transmitted to the detector 11.
0 is received consecutively. The output of the detector is thus a continuous representation of the light scattered by the cell from the minimum angle received by the detector as the cell passes into the housing to the maximum angle received as the cell exits the housing. .
This representation, when plotted as a function of time (and thus as a function of scattering angle), results in the differential light scattering pattern of the cell and can be used in the analysis system described above. As the cells approach mirror 104, they are deflected into reservoir 114 by a jet from air jet 112. Another method for measuring differential light scattering patterns is to use a single rotating detector.
If this single detector is made to rotate around the particle for a period shorter than the particle's transit time through the orthogonal illumination beam, a sufficient DLS pattern will be obtained. Such a form is described in Science Vol.
190, October 1975, pp. 375-377, in the paper entitled "Precise Measurement of Particulate Matter by Light Scattering in Optical Resonance". Still alternatively, the particles may be captured electromagnetically and scanned individually. See US Pat. No. 3,754,830. For subsequent mathematical analysis, such a DLS pattern or a portion thereof can be converted into a digital representation. As previously discussed and re-emphasized here, such a pattern has N coefficients (N2
ρ) or by N individual intensity values over the angular range of interest. For digital purposes, it is probably most economical to store such a DLS pattern for N coefficients. based on these later
DLS patterns can be recovered. This is much more economical than the equivalent number of digitally stored DLS patterns obtained from a synthetic continuum array resulting from a single detector configuration of the type described above. Although preferred systems and elements have been disclosed above, depending on the number of cells desired to be stored by the system per minute, more slow-acting and less expensive equipment may be used, and and rapid-acting equipment may also be required. Of course, the cycle time of these instruments is also related to the number of detectors in the detector array. Ten
For the system disclosed here using a detector array of ~50 detectors, cell measurement rates reach the order of 1000-60000 cells/min. This is fast enough for most cell sorting needs. In addition to using faster operating equipment, higher classification speeds can be achieved by using multiple memories and microprocessing systems. In a cell stream, the average cell velocity will be considerably smaller than the maximum cell measurement velocity due to the fact that many droplets do not contain cells and are therefore deflected from the cell stream by the cell sorter. Although preferred embodiments of this invention have been described above, it should be noted that many modifications may be made within the scope of this invention.

[Brief explanation of the drawing]

FIG. 1 is a schematic flow sheet of a preferred apparatus for use in the method of the present invention. FIG. 2 is a schematic illustration of the cell handling portion of the preferred device. FIG. 3 shows a horizontal section through the detector housing showing the detector array. Figures 4A, 4B, 4C and 4D show examples of infrared differential light scattering patterns obtained from cells with high water content. FIG. 5 is a schematic flow sheet of the analysis system. Figures 6a, 6b and 6c are similar to Figure 3 showing a modified detector arrangement. FIG. 7 shows another example of the detector configuration. 2: Sorter (classifier), 6: Detector, 8: Recorder, 22: Orifice, 26: Droplet, 30: Detector, 32: Analyzer, 34: Electrostatic deflection plate, 36:
Droplet without cells, 38: Droplet with cells, 4
4: Sensor array, 42: Housing (scattering chamber), 46: Inlet, 48: Outlet, 50: Radiation source, 52: Infrared rays, 69: Dry air source, 72: Logarithmic amplifier, 74: Peak detector , 84: multiplexer, 82: A/D converter, 76: discriminator, 7
8: Logic control unit, 86: Memory, 88: Microprocessor, 96: Mini computer, 9
0: disk data storage device, 92: video display, 94: printer.

Claims (1)

[Claims] 1. A method for analyzing cells that are in a suspension and have a size substantially larger than the wavelength of visible light, the method comprising aerosolizing a suspension of cells so that some of the cells contain the cells. separating the cell-containing droplets from the cell-free droplets; and separating the separated cells from a monochromatic color having a wavelength in the resonant region compared to the dimensions of the cells. irradiating with a beam of optical radiation; detecting the radiation intensity of a given polarization scattered by each of the irradiated cells to obtain a differential scattering pattern characterizing the size, shape, and dielectric properties of each cell; and recording the detected pattern to characterize the irradiated cells. 2 Measure the extreme values of the detected intensity pairs for each pattern, count the number of extreme values of the detected intensity pairs to give a total count for each pattern, and calculate this total count as the average dimension interval. and comparing each detected pattern with a set of reference patterns of associated average size spacing, and analyzing the detection strength by characterizing or associating each particle with a known reference pattern by this comparison. the method of. 3. The method according to item 1 above, in which the radiation used to irradiate the cells has a wavelength that does not exceed a size substantially equal to the size of the cells and is about 1/12 or more of the size of the cells. . 4 Locate the angular position of the extreme values of the pattern represented by the angular order of the detection signal for each cell to obtain one extreme value data set for each cell, and calculate the number of extreme values within the selected angular range. sort the extreme value data set into particle size groups based on the data set, normalize each extreme intensity value to the intensity value of the reference extreme value, and compare the standardized value with the reference value to determine the size of each analyzed cell. 3. The method according to item 3, wherein the detected pattern is analyzed by associating the detected pattern with one of the reference cell sets. 5. aerosolizing a suspension of cells to produce a group of droplets, some of which contain cells; separating droplets containing cells from droplets not containing cells; and separating the separated cells. irradiating with a beam of monochromatic light radiation having a wavelength in the resonant region relative to the dimensions of the cell, and detecting the radiation intensity of a given polarization scattered by each of the irradiated cells to determine the dimensions of each cell; A standard is obtained by analyzing a suspension of known cells by a method that includes obtaining a differential scattering pattern characterizing the shape, dielectric properties, and recording the detected pattern to characterize the irradiated cells. The method according to item 4 above, for obtaining a set of values. 6. The method of item 5, comprising the step of evaporating the liquid surrounding each cell droplet before sequentially irradiating the separated cells. 7. The method according to item 6 above, wherein the cells are organic cells selected from the group consisting of mammalian cells, fungal plant spores, and pollen. 8. The method of claim 5, wherein the radiation intensity is detected by elements of the detector array at a sufficient number of angular positions relative to the direction of the illumination beam. 9 The number of positions N is N[2πD _nax n _p /λ _p +
4] θ/180° where θ is the angular range covered by the detector, n _p is the refractive index of the medium in which the measurement is being made, D _nax is the largest cell or particle diameter, and λ _p is the diameter of the incident beam. The method according to the preceding item 8, which is a wavelength in a vacuum. 10 The angular positions of the N detectors of the detector array are given by the N roots of the Tievichiev polynomial T _N-1 (X), where X=θ−(θ ₁ +θ ₂ )/2/(θ ₂ − θ ₁
)/2, and θ ₁ <θ ₂ are the two limit angles that define the angular range of the pattern to be measured).